CustomerChurnPrediction

1. Objective

2. Overview

3. Data Description

3.1) Importing and merging the dataset

#Importing dataset and merging
telecom_customer_churn <- read.csv("C:/Users/Ankur/Downloads/Google/Mobile Customer Churn/telecom_customer_churn.csv")
telecom_zipcode_population <- read.csv("C:/Users/Ankur/Downloads/Google/Mobile Customer Churn/telecom_zipcode_population.csv")
mergedf <- merge(telecom_customer_churn, telecom_zipcode_population, by = "Zip.Code")

3.2) Inspecting the dataset

Summary

summary(mergedf)

##     Zip.Code         Customer.ID      Gender          Age        Married   
##  Min.   :90001   0002-ORFBO:   1   Female:3488   Min.   :19.00   No :3641  
##  1st Qu.:92101   0003-MKNFE:   1   Male  :3555   1st Qu.:32.00   Yes:3402  
##  Median :93518   0004-TLHLJ:   1                 Median :46.00             
##  Mean   :93486   0011-IGKFF:   1                 Mean   :46.51             
##  3rd Qu.:95329   0013-EXCHZ:   1                 3rd Qu.:60.00             
##  Max.   :96150   0013-MHZWF:   1                 Max.   :80.00             
##                  (Other)   :7037                                           
##  Number.of.Dependents            City         Latitude       Longitude     
##  Min.   :0.0000       Los Angeles  : 293   Min.   :32.56   Min.   :-124.3  
##  1st Qu.:0.0000       San Diego    : 285   1st Qu.:33.99   1st Qu.:-121.8  
##  Median :0.0000       San Jose     : 112   Median :36.21   Median :-119.6  
##  Mean   :0.4687       Sacramento   : 108   Mean   :36.20   Mean   :-119.8  
##  3rd Qu.:0.0000       San Francisco: 104   3rd Qu.:38.16   3rd Qu.:-118.0  
##  Max.   :9.0000       Fresno       :  61   Max.   :41.96   Max.   :-114.2  
##                       (Other)      :6080                                   
##  Number.of.Referrals Tenure.in.Months     Offer      Phone.Service
##  Min.   : 0.000      Min.   : 1.00    None   :3877   No : 682     
##  1st Qu.: 0.000      1st Qu.: 9.00    Offer A: 520   Yes:6361     
##  Median : 0.000      Median :29.00    Offer B: 824                
##  Mean   : 1.952      Mean   :32.39    Offer C: 415                
##  3rd Qu.: 3.000      3rd Qu.:55.00    Offer D: 602                
##  Max.   :11.000      Max.   :72.00    Offer E: 805                
##                                                                   
##  Avg.Monthly.Long.Distance.Charges Multiple.Lines Internet.Service
##  Min.   : 1.01                        : 682       No :1526        
##  1st Qu.:13.05                     No :3390       Yes:5517        
##  Median :25.69                     Yes:2971                       
##  Mean   :25.42                                                    
##  3rd Qu.:37.68                                                    
##  Max.   :49.99                                                    
##  NA's   :682                                                      
##      Internet.Type  Avg.Monthly.GB.Download Online.Security Online.Backup
##             :1526   Min.   : 2.00              :1526           :1526     
##  Cable      : 830   1st Qu.:13.00           No :3498        No :3088     
##  DSL        :1652   Median :21.00           Yes:2019        Yes:2429     
##  Fiber Optic:3035   Mean   :26.19                                        
##                     3rd Qu.:30.00                                        
##                     Max.   :85.00                                        
##                     NA's   :1526                                         
##  Device.Protection.Plan Premium.Tech.Support Streaming.TV Streaming.Movies
##     :1526                  :1526                :1526        :1526        
##  No :3095               No :3473             No :2810     No :2785        
##  Yes:2422               Yes:2044             Yes:2707     Yes:2732        
##                                                                           
##                                                                           
##                                                                           
##                                                                           
##  Streaming.Music Unlimited.Data           Contract    Paperless.Billing
##     :1526           :1526       Month-to-Month:3610   No :2872         
##  No :3029        No : 772       One Year      :1550   Yes:4171         
##  Yes:2488        Yes:4745       Two Year      :1883                    
##                                                                        
##                                                                        
##                                                                        
##                                                                        
##          Payment.Method Monthly.Charge   Total.Charges    Total.Refunds   
##  Bank Withdrawal:3909   Min.   :-10.00   Min.   :  18.8   Min.   : 0.000  
##  Credit Card    :2749   1st Qu.: 30.40   1st Qu.: 400.1   1st Qu.: 0.000  
##  Mailed Check   : 385   Median : 70.05   Median :1394.5   Median : 0.000  
##                         Mean   : 63.60   Mean   :2280.4   Mean   : 1.962  
##                         3rd Qu.: 89.75   3rd Qu.:3786.6   3rd Qu.: 0.000  
##                         Max.   :118.75   Max.   :8684.8   Max.   :49.790  
##                                                                           
##  Total.Extra.Data.Charges Total.Long.Distance.Charges Total.Revenue     
##  Min.   :  0.000          Min.   :   0.00             Min.   :   21.36  
##  1st Qu.:  0.000          1st Qu.:  70.55             1st Qu.:  605.61  
##  Median :  0.000          Median : 401.44             Median : 2108.64  
##  Mean   :  6.861          Mean   : 749.10             Mean   : 3034.38  
##  3rd Qu.:  0.000          3rd Qu.:1191.10             3rd Qu.: 4801.15  
##  Max.   :150.000          Max.   :3564.72             Max.   :11979.34  
##                                                                         
##  Customer.Status         Churn.Category                        Churn.Reason 
##  Churned:1869                   :5174                                :5174  
##  Joined : 454    Attitude       : 314   Competitor had better devices: 313  
##  Stayed :4720    Competitor     : 841   Competitor made better offer : 311  
##                  Dissatisfaction: 321   Attitude of support person   : 220  
##                  Other          : 182   Don't know                   : 130  
##                  Price          : 211   Competitor offered more data : 117  
##                                         (Other)                      : 778  
##    Population    
##  Min.   :    11  
##  1st Qu.:  2344  
##  Median : 17554  
##  Mean   : 22140  
##  3rd Qu.: 36125  
##  Max.   :105285  
##

Strucure

str(mergedf)

## 'data.frame':    7043 obs. of  39 variables:
##  $ Zip.Code                         : int  90001 90001 90001 90001 90002 90002 90002 90002 90003 90003 ...
##  $ Customer.ID                      : Factor w/ 7043 levels "0002-ORFBO","0003-MKNFE",..: 5376 2308 86 2253 4701 3963 3129 3634 38 6795 ...
##  $ Gender                           : Factor w/ 2 levels "Female","Male": 1 2 2 1 2 2 2 2 1 1 ...
##  $ Age                              : int  36 36 71 66 31 46 56 68 60 49 ...
##  $ Married                          : Factor w/ 2 levels "No","Yes": 2 2 2 1 2 1 1 1 2 2 ...
##  $ Number.of.Dependents             : int  0 0 0 0 0 0 0 0 0 3 ...
##  $ City                             : Factor w/ 1106 levels "Acampo","Acton",..: 555 555 555 555 555 555 555 555 555 555 ...
##  $ Latitude                         : num  34 34 34 34 33.9 ...
##  $ Longitude                        : num  -118 -118 -118 -118 -118 ...
##  $ Number.of.Referrals              : int  0 1 1 0 9 0 0 0 4 2 ...
##  $ Tenure.in.Months                 : int  1 17 69 8 58 34 13 38 59 3 ...
##  $ Offer                            : Factor w/ 6 levels "None","Offer A",..: 6 1 1 6 1 1 1 1 3 1 ...
##  $ Phone.Service                    : Factor w/ 2 levels "No","Yes": 1 1 2 2 2 2 2 2 2 2 ...
##  $ Avg.Monthly.Long.Distance.Charges: num  NA NA 18.41 5.21 18.68 ...
##  $ Multiple.Lines                   : Factor w/ 3 levels "","No","Yes": 1 1 3 2 2 2 2 3 3 2 ...
##  $ Internet.Service                 : Factor w/ 2 levels "No","Yes": 2 2 2 2 1 2 1 2 2 2 ...
##  $ Internet.Type                    : Factor w/ 4 levels "","Cable","DSL",..: 3 2 4 4 1 3 1 4 4 4 ...
##  $ Avg.Monthly.GB.Download          : int  10 10 17 8 NA 16 NA 13 14 22 ...
##  $ Online.Security                  : Factor w/ 3 levels "","No","Yes": 2 3 2 2 1 3 1 2 3 2 ...
##  $ Online.Backup                    : Factor w/ 3 levels "","No","Yes": 3 2 3 2 1 2 1 2 3 3 ...
##  $ Device.Protection.Plan           : Factor w/ 3 levels "","No","Yes": 2 3 3 3 1 3 1 2 2 2 ...
##  $ Premium.Tech.Support             : Factor w/ 3 levels "","No","Yes": 2 2 3 2 1 2 1 2 2 3 ...
##  $ Streaming.TV                     : Factor w/ 3 levels "","No","Yes": 2 2 3 3 1 2 1 3 3 2 ...
##  $ Streaming.Movies                 : Factor w/ 3 levels "","No","Yes": 2 2 3 3 1 2 1 3 2 2 ...
##  $ Streaming.Music                  : Factor w/ 3 levels "","No","Yes": 2 2 2 2 1 2 1 2 2 2 ...
##  $ Unlimited.Data                   : Factor w/ 3 levels "","No","Yes": 3 3 3 3 1 3 1 3 3 3 ...
##  $ Contract                         : Factor w/ 3 levels "Month-to-Month",..: 1 1 3 1 3 2 1 3 1 1 ...
##  $ Paperless.Billing                : Factor w/ 2 levels "No","Yes": 2 1 2 2 1 1 2 1 2 2 ...
##  $ Payment.Method                   : Factor w/ 3 levels "Bank Withdrawal",..: 1 3 1 2 2 3 1 1 1 1 ...
##  $ Monthly.Charge                   : num  29.9 34.4 110 94.5 -4 ...
##  $ Total.Charges                    : num  29.9 592.8 7634.2 743 1186 ...
##  $ Total.Refunds                    : num  0 0 0 0 0 ...
##  $ Total.Extra.Data.Charges         : int  0 0 0 0 0 0 0 0 0 0 ...
##  $ Total.Long.Distance.Charges      : num  0 0 1270.3 41.7 1083.4 ...
##  $ Total.Revenue                    : num  29.9 592.8 8904.5 784.6 2269.4 ...
##  $ Customer.Status                  : Factor w/ 3 levels "Churned","Joined",..: 2 3 3 1 3 3 3 3 3 2 ...
##  $ Churn.Category                   : Factor w/ 6 levels "","Attitude",..: 1 1 1 3 1 1 1 1 1 1 ...
##  $ Churn.Reason                     : Factor w/ 21 levels "","Attitude of service provider",..: 1 1 1 4 1 1 1 1 1 1 ...
##  $ Population                       : int  54492 54492 54492 54492 44586 44586 44586 44586 58198 58198 ...

Head

knitr::kable(head(mergedf))

Zip.Code	Customer.ID	Gender	Age	Married	City	Latitude	Longitude	Number.of.Referrals	Tenure.in.Months	Offer	Phone.Service	Avg.Monthly.Long.Distance.Charges	Multiple.Lines	Internet.Service	Internet.Type	Avg.Monthly.GB.Download	Online.Security	Online.Backup	Device.Protection.Plan	Premium.Tech.Support	Streaming.TV	Streaming.Movies	Streaming.Music	Unlimited.Data	Contract	Paperless.Billing	Payment.Method	Monthly.Charge	Total.Charges	Total.Long.Distance.Charges	Total.Revenue	Customer.Status	Churn.Category	Churn.Reason	Population
90001	7590-VHVEG	Female	36	Yes	Los Angeles	33.97362	-118.249	0	1	Offer E	No	NA		Yes	DSL	10	No	Yes	No	No	No	No	No	Yes	Month-to-Month	Yes	Bank Withdrawal	29.85	29.85	0.00	29.85	Joined			54492
90001	3307-TLCUD	Male	36	Yes	Los Angeles	33.97362	-118.249	1	17	None	No	NA		Yes	Cable	10	Yes	No	Yes	No	No	No	No	Yes	Month-to-Month	No	Mailed Check	34.40	592.75	0.00	592.75	Stayed			54492
90001	0136-IFMYD	Male	71	Yes	Los Angeles	33.97362	-118.249	1	69	None	Yes	18.41	Yes	Yes	Fiber Optic	17	No	Yes	Yes	Yes	Yes	Yes	No	Yes	Two Year	Yes	Bank Withdrawal	109.95	7634.25	1270.29	8904.54	Stayed			54492
90001	3217-FZDMN	Female	66	No	Los Angeles	33.97362	-118.249	0	8	Offer E	Yes	5.21	No	Yes	Fiber Optic	8	No	No	Yes	No	Yes	Yes	No	Yes	Month-to-Month	Yes	Credit Card	94.45	742.95	41.68	784.63	Churned	Competitor	Competitor had better devices	54492
90002	6625-FLENO	Male	31	Yes	Los Angeles	33.94926	-118.247	9	58	None	Yes	18.68	No	No		NA									Two Year	No	Credit Card	-4.00	1185.95	1083.44	2269.39	Stayed			44586
90002	5575-GNVDE	Male	46	No	Los Angeles	33.94926	-118.247	0	34	None	Yes	17.09	No	Yes	DSL	16	Yes	No	Yes	No	No	No	No	Yes	One Year	No	Mailed Check	56.95	1889.50	581.06	2470.56	Stayed			44586

4. Map View

Map view of all observation

library(mapview)
mapview(telecom_customer_churn, xcol = "Longitude", ycol = "Latitude", crs = 4269, grid = FALSE)

Count of users in each ZipCode

library(tidyverse)
library(dplyr)
library(sf)
#Grouped by ZipCode and finding out count of users in these areas
df <- mergedf %>%
  group_by(Zip.Code) %>%
  count(Zip.Code, name = "Count", sort = TRUE, Longitude, Latitude)
#Converting int/numerical type coordinate to sfc_POINT(geometry)
df1 <- st_as_sf(df, coords = c("Longitude", "Latitude"), crs= 4269)
mapview(df1, zcol = "Count", at = seq(0, 45, 5), alpha = 0.2, legend = TRUE) #Count of users in mapview

Percentage of User Churned in specific ZipCode

#Converting int/numerical type coordinate to sfc_POINT(geometry)
df1 <- st_as_sf(df, coords = c("Longitude", "Latitude"), crs= 4269)
#Percentage of User Churned in specific ZipCode
churned_cust <- mergedf %>%
  filter(Customer.Status == "Churned")
churned_cust <- churned_cust %>%
  group_by(Zip.Code) %>%
  count(Zip.Code, name = "Churn_Count", sort = TRUE, Longitude, Latitude)
churned_cust <- merge(df1, churned_cust, by = "Zip.Code")
churned_cust$Percent.Churn <- (churned_cust$Churn_Count/churned_cust$Count)*100
mapview(churned_cust, zcol = "Percent.Churn", at = seq(0,100,25), legend = TRUE)

5. Data Cleaning and manipulation

mergedf3 <- mergedf
mergedf3[is.na(mergedf3)] = 0 #Replacing NA values with 0 from avg.monthly charges and avg.monthly GB used 
mergedf3$Multiple.Lines[mergedf3$Multiple.Lines == ""] <- "No" #Replacing blank values with "No"
transformedf <- mergedf3
transformedf$Offer <- recode_factor(transformedf$Offer, "None" = 0, "Offer A" = 1, "Offer B" = 2, "Offer C" = 3, "Offer D" = 4, "Offer E" = 5)
transformedf$Phone.Service <- recode_factor(transformedf$Phone.Service, No = 0, Yes = 1)
transformedf$Multiple.Lines <- recode_factor(transformedf$Multiple.Lines, No = 0, Yes = 1)
transformedf$Internet.Service <- recode_factor(transformedf$Internet.Service, No = 0, Yes =1)
transformedf$Internet.Type <- recode(transformedf$Internet.Type, "Cable" = 1, "DSL" = 2, "Fiber Optic" = 3)
transformedf$Internet.Type <- recode_factor(transformedf$Internet.Type, "Cable" = 1, "DSL" = 2, "Fiber Optic" = 3, .missing = 0)
transformedf$Online.Security <- recode(transformedf$Online.Security, No = 0, Yes = 1) #Unmatched values(Blanks in here) are coverted to NAs
transformedf$Online.Security <- recode_factor(transformedf$Online.Security, No = 0, Yes = 1, .missing = 0) #NAs are recoded as 0
transformedf$Online.Backup <- recode(transformedf$Online.Backup, No = 0, Yes = 1)
transformedf$Online.Backup <- recode_factor(transformedf$Online.Backup, No = 0, Yes = 1, .missing = 0)
transformedf$Device.Protection.Plan <- recode(transformedf$Device.Protection.Plan, No = 0, Yes = 1)
transformedf$Device.Protection.Plan <- recode_factor(transformedf$Device.Protection.Plan, No = 0, Yes = 1, .missing = 0)
transformedf$Premium.Tech.Support <- recode(transformedf$Premium.Tech.Support, No = 0, Yes = 1)
transformedf$Premium.Tech.Support <- recode_factor(transformedf$Premium.Tech.Support, No = 0, Yes = 1, .missing = 0)
transformedf$Streaming.TV <- recode(transformedf$Streaming.TV, No = 0, Yes = 1)
transformedf$Streaming.TV <- recode_factor(transformedf$Streaming.TV, No = 0, Yes = 1, .missing = 0)
transformedf$Streaming.Movies <- recode(transformedf$Streaming.Movies, No = 0, Yes = 1)
transformedf$Streaming.Movies <- recode_factor(transformedf$Streaming.Movies, No = 0, Yes = 1, .missing = 0)
transformedf$Streaming.Music <- recode(transformedf$Streaming.Music, No = 0, Yes = 1)
transformedf$Streaming.Music <- recode_factor(transformedf$Streaming.Music, No = 0, Yes = 1, .missing = 0)
transformedf$Unlimited.Data <- recode(transformedf$Unlimited.Data, No = 0, Yes = 1)
transformedf$Unlimited.Data <- recode_factor(transformedf$Unlimited.Data, No = 0, Yes = 1, .missing = 0)
transformedf$Contract <- recode_factor(transformedf$Contract, "Month-to-Month" = 1, "One Year" = 2, "Two Year" = 3)
transformedf$Paperless.Billing <- recode_factor(transformedf$Paperless.Billing, No = 0, Yes = 1)
transformedf$Payment.Method <- recode_factor(transformedf$Payment.Method, "Bank Withdrawal" = 1, "Credit Card" = 2, "Mailed Check" = 3)
transformedf$Churn.Category <- recode(transformedf$Churn.Category, "Attitude" = 1, "Competitor" = 2, "Dissatisfaction" = 3, "Price" = 4, "Other" = 5)
transformedf$Churn.Category <- recode_factor(transformedf$Churn.Category, "Attitude" = 1, "Competitor" = 2, "Dissatisfaction" = 3, "Price" = 4, "Other" = 5, .missing = 0)

Summary

summary(transformedf)

##     Zip.Code         Customer.ID      Gender          Age        Married   
##  Min.   :90001   0002-ORFBO:   1   Female:3488   Min.   :19.00   No :3641  
##  1st Qu.:92101   0003-MKNFE:   1   Male  :3555   1st Qu.:32.00   Yes:3402  
##  Median :93518   0004-TLHLJ:   1                 Median :46.00             
##  Mean   :93486   0011-IGKFF:   1                 Mean   :46.51             
##  3rd Qu.:95329   0013-EXCHZ:   1                 3rd Qu.:60.00             
##  Max.   :96150   0013-MHZWF:   1                 Max.   :80.00             
##                  (Other)   :7037                                           
##  Number.of.Dependents            City         Latitude       Longitude     
##  Min.   :0.0000       Los Angeles  : 293   Min.   :32.56   Min.   :-124.3  
##  1st Qu.:0.0000       San Diego    : 285   1st Qu.:33.99   1st Qu.:-121.8  
##  Median :0.0000       San Jose     : 112   Median :36.21   Median :-119.6  
##  Mean   :0.4687       Sacramento   : 108   Mean   :36.20   Mean   :-119.8  
##  3rd Qu.:0.0000       San Francisco: 104   3rd Qu.:38.16   3rd Qu.:-118.0  
##  Max.   :9.0000       Fresno       :  61   Max.   :41.96   Max.   :-114.2  
##                       (Other)      :6080                                   
##  Number.of.Referrals Tenure.in.Months Offer    Phone.Service
##  Min.   : 0.000      Min.   : 1.00    0:3877   0: 682       
##  1st Qu.: 0.000      1st Qu.: 9.00    1: 520   1:6361       
##  Median : 0.000      Median :29.00    2: 824                
##  Mean   : 1.952      Mean   :32.39    3: 415                
##  3rd Qu.: 3.000      3rd Qu.:55.00    4: 602                
##  Max.   :11.000      Max.   :72.00    5: 805                
##                                                             
##  Avg.Monthly.Long.Distance.Charges Multiple.Lines Internet.Service
##  Min.   : 0.00                     0:4072         0:1526          
##  1st Qu.: 9.21                     1:2971         1:5517          
##  Median :22.89                                                    
##  Mean   :22.96                                                    
##  3rd Qu.:36.40                                                    
##  Max.   :49.99                                                    
##                                                                   
##  Internet.Type Avg.Monthly.GB.Download Online.Security Online.Backup
##  1: 830        Min.   : 0.00           0:5024          0:4614       
##  2:1652        1st Qu.: 3.00           1:2019          1:2429       
##  3:3035        Median :17.00                                        
##  0:1526        Mean   :20.52                                        
##                3rd Qu.:27.00                                        
##                Max.   :85.00                                        
##                                                                     
##  Device.Protection.Plan Premium.Tech.Support Streaming.TV Streaming.Movies
##  0:4621                 0:4999               0:4336       0:4311          
##  1:2422                 1:2044               1:2707       1:2732          
##                                                                           
##                                                                           
##                                                                           
##                                                                           
##                                                                           
##  Streaming.Music Unlimited.Data Contract Paperless.Billing Payment.Method
##  0:4555          0:2298         1:3610   0:2872            1:3909        
##  1:2488          1:4745         2:1550   1:4171            2:2749        
##                                 3:1883                     3: 385        
##                                                                          
##                                                                          
##                                                                          
##                                                                          
##  Monthly.Charge   Total.Charges    Total.Refunds    Total.Extra.Data.Charges
##  Min.   :-10.00   Min.   :  18.8   Min.   : 0.000   Min.   :  0.000         
##  1st Qu.: 30.40   1st Qu.: 400.1   1st Qu.: 0.000   1st Qu.:  0.000         
##  Median : 70.05   Median :1394.5   Median : 0.000   Median :  0.000         
##  Mean   : 63.60   Mean   :2280.4   Mean   : 1.962   Mean   :  6.861         
##  3rd Qu.: 89.75   3rd Qu.:3786.6   3rd Qu.: 0.000   3rd Qu.:  0.000         
##  Max.   :118.75   Max.   :8684.8   Max.   :49.790   Max.   :150.000         
##                                                                             
##  Total.Long.Distance.Charges Total.Revenue      Customer.Status Churn.Category
##  Min.   :   0.00             Min.   :   21.36   Churned:1869    1: 314        
##  1st Qu.:  70.55             1st Qu.:  605.61   Joined : 454    2: 841        
##  Median : 401.44             Median : 2108.64   Stayed :4720    3: 321        
##  Mean   : 749.10             Mean   : 3034.38                   4: 211        
##  3rd Qu.:1191.10             3rd Qu.: 4801.15                   5: 182        
##  Max.   :3564.72             Max.   :11979.34                   0:5174        
##                                                                               
##                         Churn.Reason    Population    
##                               :5174   Min.   :    11  
##  Competitor had better devices: 313   1st Qu.:  2344  
##  Competitor made better offer : 311   Median : 17554  
##  Attitude of support person   : 220   Mean   : 22140  
##  Don't know                   : 130   3rd Qu.: 36125  
##  Competitor offered more data : 117   Max.   :105285  
##  (Other)                      : 778

Structure

str(transformedf)

## 'data.frame':    7043 obs. of  39 variables:
##  $ Zip.Code                         : int  90001 90001 90001 90001 90002 90002 90002 90002 90003 90003 ...
##  $ Customer.ID                      : Factor w/ 7043 levels "0002-ORFBO","0003-MKNFE",..: 5376 2308 86 2253 4701 3963 3129 3634 38 6795 ...
##  $ Gender                           : Factor w/ 2 levels "Female","Male": 1 2 2 1 2 2 2 2 1 1 ...
##  $ Age                              : int  36 36 71 66 31 46 56 68 60 49 ...
##  $ Married                          : Factor w/ 2 levels "No","Yes": 2 2 2 1 2 1 1 1 2 2 ...
##  $ Number.of.Dependents             : int  0 0 0 0 0 0 0 0 0 3 ...
##  $ City                             : Factor w/ 1106 levels "Acampo","Acton",..: 555 555 555 555 555 555 555 555 555 555 ...
##  $ Latitude                         : num  34 34 34 34 33.9 ...
##  $ Longitude                        : num  -118 -118 -118 -118 -118 ...
##  $ Number.of.Referrals              : int  0 1 1 0 9 0 0 0 4 2 ...
##  $ Tenure.in.Months                 : int  1 17 69 8 58 34 13 38 59 3 ...
##  $ Offer                            : Factor w/ 6 levels "0","1","2","3",..: 6 1 1 6 1 1 1 1 3 1 ...
##  $ Phone.Service                    : Factor w/ 2 levels "0","1": 1 1 2 2 2 2 2 2 2 2 ...
##  $ Avg.Monthly.Long.Distance.Charges: num  0 0 18.41 5.21 18.68 ...
##  $ Multiple.Lines                   : Factor w/ 2 levels "0","1": 1 1 2 1 1 1 1 2 2 1 ...
##  $ Internet.Service                 : Factor w/ 2 levels "0","1": 2 2 2 2 1 2 1 2 2 2 ...
##  $ Internet.Type                    : Factor w/ 4 levels "1","2","3","0": 2 1 3 3 4 2 4 3 3 3 ...
##  $ Avg.Monthly.GB.Download          : num  10 10 17 8 0 16 0 13 14 22 ...
##  $ Online.Security                  : Factor w/ 2 levels "0","1": 1 2 1 1 1 2 1 1 2 1 ...
##  $ Online.Backup                    : Factor w/ 2 levels "0","1": 2 1 2 1 1 1 1 1 2 2 ...
##  $ Device.Protection.Plan           : Factor w/ 2 levels "0","1": 1 2 2 2 1 2 1 1 1 1 ...
##  $ Premium.Tech.Support             : Factor w/ 2 levels "0","1": 1 1 2 1 1 1 1 1 1 2 ...
##  $ Streaming.TV                     : Factor w/ 2 levels "0","1": 1 1 2 2 1 1 1 2 2 1 ...
##  $ Streaming.Movies                 : Factor w/ 2 levels "0","1": 1 1 2 2 1 1 1 2 1 1 ...
##  $ Streaming.Music                  : Factor w/ 2 levels "0","1": 1 1 1 1 1 1 1 1 1 1 ...
##  $ Unlimited.Data                   : Factor w/ 2 levels "0","1": 2 2 2 2 1 2 1 2 2 2 ...
##  $ Contract                         : Factor w/ 3 levels "1","2","3": 1 1 3 1 3 2 1 3 1 1 ...
##  $ Paperless.Billing                : Factor w/ 2 levels "0","1": 2 1 2 2 1 1 2 1 2 2 ...
##  $ Payment.Method                   : Factor w/ 3 levels "1","2","3": 1 3 1 2 2 3 1 1 1 1 ...
##  $ Monthly.Charge                   : num  29.9 34.4 110 94.5 -4 ...
##  $ Total.Charges                    : num  29.9 592.8 7634.2 743 1186 ...
##  $ Total.Refunds                    : num  0 0 0 0 0 ...
##  $ Total.Extra.Data.Charges         : int  0 0 0 0 0 0 0 0 0 0 ...
##  $ Total.Long.Distance.Charges      : num  0 0 1270.3 41.7 1083.4 ...
##  $ Total.Revenue                    : num  29.9 592.8 8904.5 784.6 2269.4 ...
##  $ Customer.Status                  : Factor w/ 3 levels "Churned","Joined",..: 2 3 3 1 3 3 3 3 3 2 ...
##  $ Churn.Category                   : Factor w/ 6 levels "1","2","3","4",..: 6 6 6 2 6 6 6 6 6 6 ...
##  $ Churn.Reason                     : Factor w/ 21 levels "","Attitude of service provider",..: 1 1 1 4 1 1 1 1 1 1 ...
##  $ Population                       : int  54492 54492 54492 54492 44586 44586 44586 44586 58198 58198 ...

Head

knitr::kable(head(transformedf))

Zip.Code	Customer.ID	Gender	Age	Married	City	Latitude	Longitude	Number.of.Referrals	Tenure.in.Months	Offer	Phone.Service	Avg.Monthly.Long.Distance.Charges	Multiple.Lines	Internet.Service	Internet.Type	Avg.Monthly.GB.Download	Online.Security	Online.Backup	Device.Protection.Plan	Premium.Tech.Support	Streaming.TV	Streaming.Movies	Unlimited.Data	Contract	Paperless.Billing	Payment.Method	Monthly.Charge	Total.Charges	Total.Long.Distance.Charges	Total.Revenue	Customer.Status	Churn.Category	Churn.Reason	Population
90001	7590-VHVEG	Female	36	Yes	Los Angeles	33.97362	-118.249	0	1	5	0	0.00	0	1	2	10	0	1	0	0	0	0	1	1	1	1	29.85	29.85	0.00	29.85	Joined	0		54492
90001	3307-TLCUD	Male	36	Yes	Los Angeles	33.97362	-118.249	1	17	0	0	0.00	0	1	1	10	1	0	1	0	0	0	1	1	0	3	34.40	592.75	0.00	592.75	Stayed	0		54492
90001	0136-IFMYD	Male	71	Yes	Los Angeles	33.97362	-118.249	1	69	0	1	18.41	1	1	3	17	0	1	1	1	1	1	1	3	1	1	109.95	7634.25	1270.29	8904.54	Stayed	0		54492
90001	3217-FZDMN	Female	66	No	Los Angeles	33.97362	-118.249	0	8	5	1	5.21	0	1	3	8	0	0	1	0	1	1	1	1	1	2	94.45	742.95	41.68	784.63	Churned	2	Competitor had better devices	54492
90002	6625-FLENO	Male	31	Yes	Los Angeles	33.94926	-118.247	9	58	0	1	18.68	0	0	0	0	0	0	0	0	0	0	0	3	0	2	-4.00	1185.95	1083.44	2269.39	Stayed	0		44586
90002	5575-GNVDE	Male	46	No	Los Angeles	33.94926	-118.247	0	34	0	1	17.09	0	1	2	16	1	0	1	0	0	0	1	2	0	3	56.95	1889.50	581.06	2470.56	Stayed	0		44586

6. Data Split

#Dividing the dataset into Train & Test dataset
set.seed(123) #For the same output, we will utilize set.seed function
transformedf <- transformedf
split1 <- sample(c(rep(0, 0.7*nrow(transformedf)), rep(1, 0.3*nrow(transformedf))))
table(split1)

## split1
##    0    1 
## 4930 2112

traindf <- transformedf[split1 == 0,]
testdf <- transformedf[split1 == 1,]

Train Dataset

summary(traindf$Customer.Status)

## Churned  Joined  Stayed 
##    1325     327    3279

Test Dataset

summary(testdf$Customer.Status)

## Churned  Joined  Stayed 
##     544     127    1441

7. Preliminary Data Analysis

Chi-square and Cramer’s V

library(gsubfn)
#Chi-square statistics, p-value and Cramer's V value of categorical variables
traindf_cat <- traindf[,c("Offer", "Phone.Service", "Internet.Service", "Internet.Type",
                          "Online.Security", "Online.Backup", "Device.Protection.Plan",
                          "Premium.Tech.Support", "Streaming.TV", "Streaming.Movies","Streaming.Music", "Unlimited.Data",
                          "Contract", "Paperless.Billing", "Payment.Method", "Customer.Status")] 
CHIS <- lapply(traindf_cat[,-16], function(x) chisq.test(traindf_cat[,16], x));
CHIs_table <- do.call(rbind, CHIS)[,c(1,3)]
CHIs_table <- as.data.frame(CHIs_table)
CHIs_table$statistic <- gsub("c", "", as.numeric(CHIs_table$statistic))
CHIs_table$p.value <- as.numeric(CHIs_table$p.value)
d <- CHIs_table
names <- rownames(d)
CHIs_table$Variable <- names
level <- traindf_cat %>%
  summarise_all(n_distinct)
level <- as.data.frame(t(level))
names <- rownames(level)
level$Variable <- names
CHIs_table <- merge(CHIs_table, level, by = "Variable")
CHIs_table$statistic <- as.numeric(CHIs_table$statistic)
CHIs_table$Cramers_Value <- sqrt(((CHIs_table$statistic)/nrow(traindf))/min(CHIs_table$V1, n_distinct(traindf$Customer.Status)))
knitr::kable(CHIs_table)

Variable	statistic	p.value	V1	Cramers_Value
Contract	1441.3859	0.0000000	3	0.3823029
Device.Protection.Plan	156.9618	0.0000000	2	0.1261579
Internet.Service	309.0694	0.0000000	2	0.1770294
Internet.Type	532.7079	0.0000000	4	0.2324139
Offer	795.4861	0.0000000	6	0.2840101
Online.Backup	138.4891	0.0000000	2	0.1185019
Online.Security	227.8757	0.0000000	2	0.1520080
Paperless.Billing	205.1861	0.0000000	2	0.1442419
Payment.Method	258.3974	0.0000000	3	0.1618682
Phone.Service	3.8192	0.1481396	2	0.0196790
Premium.Tech.Support	245.9676	0.0000000	2	0.1579270
Streaming.Movies	133.0671	0.0000000	2	0.1161590
Streaming.Music	102.2052	0.0000000	2	0.1018015
Streaming.TV	133.7242	0.0000000	2	0.1164455
Unlimited.Data	168.5999	0.0000000	2	0.1307513

Boxplot

traindf1 <- traindf[,c(4,6,10,11:36)]
cat <- traindf1[,c(1:4,7,11,23:28,29)]
library(ggpubr)

MANOVA

library(nnet)
res.man <- manova(cbind(Age, Number.of.Dependents,Number.of.Referrals, Tenure.in.Months, Avg.Monthly.Long.Distance.Charges, Avg.Monthly.GB.Download, Total.Extra.Data.Charges, Total.Long.Distance.Charges, Total.Revenue, Monthly.Charge, Total.Charges, Total.Refunds) ~ Customer.Status, data = cat)
summary(res.man, tol = 0)

##                   Df  Pillai approx F num Df den Df    Pr(>F)    
## Customer.Status    2 0.45123    119.4     24   9836 < 2.2e-16 ***
## Residuals       4928                                             
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1

summary.aov(res.man)

##  Response Age :
##                   Df  Sum Sq Mean Sq F value    Pr(>F)    
## Customer.Status    2   19843  9921.3    35.7 4.047e-16 ***
## Residuals       4928 1369541   277.9                      
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
##  Response Number.of.Dependents :
##                   Df Sum Sq Mean Sq F value    Pr(>F)    
## Customer.Status    2  238.5 119.245  139.05 < 2.2e-16 ***
## Residuals       4928 4226.1   0.858                      
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
##  Response Number.of.Referrals :
##                   Df Sum Sq Mean Sq F value    Pr(>F)    
## Customer.Status    2   4209 2104.49  264.36 < 2.2e-16 ***
## Residuals       4928  39231    7.96                      
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
##  Response Tenure.in.Months :
##                   Df  Sum Sq Mean Sq F value    Pr(>F)    
## Customer.Status    2  848588  424294  982.02 < 2.2e-16 ***
## Residuals       4928 2129193     432                      
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
##  Response Avg.Monthly.Long.Distance.Charges :
##                   Df  Sum Sq Mean Sq F value Pr(>F)
## Customer.Status    2     322  161.01  0.6742 0.5096
## Residuals       4928 1176863  238.81               
## 
##  Response Avg.Monthly.GB.Download :
##                   Df  Sum Sq Mean Sq F value    Pr(>F)    
## Customer.Status    2   15492  7746.2  18.794 7.395e-09 ***
## Residuals       4928 2031164   412.2                      
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
##  Response Total.Extra.Data.Charges :
##                   Df  Sum Sq Mean Sq F value    Pr(>F)    
## Customer.Status    2    9543  4771.4  7.4154 0.0006086 ***
## Residuals       4928 3170873   643.4                      
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
##  Response Total.Long.Distance.Charges :
##                   Df     Sum Sq   Mean Sq F value    Pr(>F)    
## Customer.Status    2  425524334 212762167  335.61 < 2.2e-16 ***
## Residuals       4928 3124150840    633959                      
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
##  Response Total.Revenue :
##                   Df     Sum Sq    Mean Sq F value    Pr(>F)    
## Customer.Status    2 6.0482e+09 3024094638  432.22 < 2.2e-16 ***
## Residuals       4928 3.4480e+10    6996664                      
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
##  Response Monthly.Charge :
##                   Df  Sum Sq Mean Sq F value    Pr(>F)    
## Customer.Status    2  292149  146074  159.66 < 2.2e-16 ***
## Residuals       4928 4508547     915                      
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
##  Response Total.Charges :
##                   Df     Sum Sq    Mean Sq F value    Pr(>F)    
## Customer.Status    2 3.2710e+09 1635484274  363.77 < 2.2e-16 ***
## Residuals       4928 2.2156e+10    4495902                      
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
##  Response Total.Refunds :
##                   Df Sum Sq Mean Sq F value    Pr(>F)    
## Customer.Status    2   1555  777.60   13.11 2.097e-06 ***
## Residuals       4928 292304   59.32                      
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1

Correlation

knitr::kable(cor(cat[1:12]))

	Age	Number.of.Dependents	Number.of.Referrals	Tenure.in.Months	Avg.Monthly.Long.Distance.Charges	Avg.Monthly.GB.Download	Monthly.Charge	Total.Charges	Total.Refunds	Total.Extra.Data.Charges	Total.Long.Distance.Charges	Total.Revenue
Age	1.0000000	-0.1233383	-0.0310285	0.0085981	-0.0135942	-0.3724461	0.1297730	0.0536561	0.0201893	0.0099749	-0.0056391	0.0408650
Number.of.Dependents	-0.1233383	1.0000000	0.2865629	0.1132867	-0.0111846	0.1426338	-0.1217353	0.0249403	0.0120779	-0.0196748	0.0686656	0.0398695
Number.of.Referrals	-0.0310285	0.2865629	1.0000000	0.3153105	0.0063733	0.0384084	0.0200994	0.2388944	0.0165633	-0.0052310	0.2115740	0.2517480
Tenure.in.Months	0.0085981	0.1132867	0.3153105	1.0000000	0.0083527	0.0518238	0.2379141	0.8251031	0.0622769	0.0868819	0.6719726	0.8530203
Avg.Monthly.Long.Distance.Charges	-0.0135942	-0.0111846	0.0063733	0.0083527	1.0000000	-0.0261397	0.1306104	0.0638032	-0.0089866	0.0022397	0.5989450	0.2278390
Avg.Monthly.GB.Download	-0.3724461	0.1426338	0.0384084	0.0518238	-0.0261397	1.0000000	0.3713364	0.2285263	0.0050984	0.0907955	0.0170669	0.1868528
Monthly.Charge	0.1297730	-0.1217353	0.0200994	0.2379141	0.1306104	0.3713364	1.0000000	0.6254960	0.0210074	0.1150761	0.2328548	0.5653195
Total.Charges	0.0536561	0.0249403	0.2388944	0.8251031	0.0638032	0.2285263	0.6254960	1.0000000	0.0403317	0.1197636	0.6039930	0.9717846
Total.Refunds	0.0201893	0.0120779	0.0165633	0.0622769	-0.0089866	0.0050984	0.0210074	0.0403317	1.0000000	0.0022219	0.0367774	0.0401572
Total.Extra.Data.Charges	0.0099749	-0.0196748	-0.0052310	0.0868819	0.0022397	0.0907955	0.1150761	0.1197636	0.0022219	1.0000000	0.0600591	0.1214895
Total.Long.Distance.Charges	-0.0056391	0.0686656	0.2115740	0.6719726	0.5989450	0.0170669	0.2328548	0.6039930	0.0367774	0.0600591	1.0000000	0.7747940
Total.Revenue	0.0408650	0.0398695	0.2517480	0.8530203	0.2278390	0.1868528	0.5653195	0.9717846	0.0401572	0.1214895	0.7747940	1.0000000

8. Multinomial Logistic Regression

Model Summary

summary(zz2)

## Call:
## multinom(formula = Customer.Status ~ Age + Number.of.Dependents + 
##     Number.of.Referrals + Tenure.in.Months + Offer + Phone.Service + 
##     Avg.Monthly.Long.Distance.Charges + Multiple.Lines + Internet.Service + 
##     Internet.Type + Avg.Monthly.GB.Download + Online.Security + 
##     Online.Backup + Device.Protection.Plan + Premium.Tech.Support + 
##     Streaming.TV + Streaming.Movies + Streaming.Music + Unlimited.Data + 
##     Contract + Paperless.Billing + Payment.Method + Monthly.Charge + 
##     Total.Charges + Total.Refunds + Total.Long.Distance.Charges + 
##     Total.Revenue, data = traindf_pred2)
## 
## Coefficients:
##        (Intercept)         Age Number.of.Dependents Number.of.Referrals
## Joined   1.9603511 -0.03082223            0.3205773           0.3305210
## Stayed  -0.3169966 -0.01679600            0.5478115           0.2502277
##        Tenure.in.Months    Offer1     Offer2    Offer3     Offer4     Offer5
## Joined      -0.95900384 23.549183 10.1865275 0.5313735 -8.1349470 -0.3594837
## Stayed       0.09250368 -1.287368  0.1061509 0.2216949  0.7543357 -0.3631852
##        Phone.Service1 Avg.Monthly.Long.Distance.Charges Multiple.Lines1
## Joined      0.5212698                      -0.004649323      -0.5270837
## Stayed      0.6956961                       0.003546937      -0.1104598
##        Internet.Service1 Internet.Type2 Internet.Type3 Internet.Type0
## Joined         0.4653483      0.3908882     -0.6612113      1.4950027
## Stayed        -0.5381925      0.4673270      0.3413906      0.2211959
##        Avg.Monthly.GB.Download Online.Security1 Online.Backup1
## Joined            -0.002371780        0.5702509      0.6440871
## Stayed            -0.004288278        0.5487430      0.3263126
##        Device.Protection.Plan1 Premium.Tech.Support1 Streaming.TV1
## Joined              0.06356905             0.5755499    -0.2902495
## Stayed              0.14652193             0.5452480    -0.1623621
##        Streaming.Movies1 Streaming.Music1 Unlimited.Data1 Contract2 Contract3
## Joined         1.3570234       -1.7336383     -0.33594223  2.107196  3.630894
## Stayed         0.1798889       -0.1109168      0.02103502  1.351127  2.872753
##        Paperless.Billing1 Payment.Method2 Payment.Method3 Monthly.Charge
## Joined         -0.4922908       0.6859813      -0.5014792   -0.002071048
## Stayed         -0.2423102       0.4159249      -0.5312636   -0.011891290
##        Total.Charges Total.Refunds Total.Long.Distance.Charges Total.Revenue
## Joined  0.0031422737   0.009208220               -0.0010371218  0.0007649897
## Stayed -0.0003954708   0.006072633                0.0002451031 -0.0003175382
## 
## Std. Errors:
##        (Intercept)         Age Number.of.Dependents Number.of.Referrals
## Joined 0.001706426 0.003754653            0.0218358          0.04522931
## Stayed 0.002729437 0.002442088            0.0565553          0.02474783
##        Tenure.in.Months       Offer1       Offer2       Offer3       Offer4
## Joined      0.007100867 8.104668e-06 4.419837e-07 3.205531e-08 3.841388e-09
## Stayed      0.007684787 2.729151e-04 3.213364e-04 1.162494e-04 2.556609e-04
##              Offer5 Phone.Service1 Avg.Monthly.Long.Distance.Charges
## Joined 0.0006479004    0.001770351                       0.007799667
## Stayed 0.0010671972    0.003084193                       0.004258883
##        Multiple.Lines1 Internet.Service1 Internet.Type2 Internet.Type3
## Joined    0.0005638495       0.001570233   0.0005032142    0.001177688
## Stayed    0.0003208895       0.002722709   0.0005540282    0.003801231
##        Internet.Type0 Avg.Monthly.GB.Download Online.Security1 Online.Backup1
## Joined    0.003211291             0.004674395     0.0004801799   0.0005627926
## Stayed    0.005395436             0.002513387     0.0016159531   0.0010616302
##        Device.Protection.Plan1 Premium.Tech.Support1 Streaming.TV1
## Joined            0.0002420052          0.0004836781  0.0003142431
## Stayed            0.0009277510          0.0015734487  0.0007492829
##        Streaming.Movies1 Streaming.Music1 Unlimited.Data1    Contract2
## Joined      0.0004481977     0.0002423772     0.001871512 0.0002026849
## Stayed      0.0007094045     0.0007221277     0.003129928 0.0003659569
##           Contract3 Paperless.Billing1 Payment.Method2 Payment.Method3
## Joined 0.0002067809       0.0009340948     0.002852559    0.0002641255
## Stayed 0.0004743466       0.0031277831     0.005896004    0.0005684454
##        Monthly.Charge Total.Charges Total.Refunds Total.Long.Distance.Charges
## Joined    0.004396115   0.009263191   0.023705076                 0.009740034
## Stayed    0.002213205   0.001630616   0.006066285                 0.001645722
##        Total.Revenue
## Joined   0.009209219
## Stayed   0.001632669
## 
## Residual Deviance: 4044.068 
## AIC: 4184.068

Model Fit in Train Dataset & Accuracy

#Fitting model2 in train dataset
traindf_pred2$predicted <- predict(zz2,newdata = traindf_pred2, traindf_pred2$predicted)
tab_train2 <- table(traindf_pred2$Customer.Status, traindf_pred2$predicted)
knitr::kable(tab_train2)

	Churned	Joined	Stayed
Churned	887	110	328
Joined	95	221	11
Stayed	329	3	2947

# Calculating accuracy - sum of diagonal elements divided by total obs
round((sum(diag(tab_train2))/sum(tab_train2))*100,2)

## [1] 82.23

Model Fit in Test Dataset & Accuracy

#Fitting the model in test dataframe
test_pred2 <- testdf
test_pred2$predicted <- predict(zz2,newdata = test_pred2, test_pred2$predicted)
tab_test2 <- table(test_pred2$Customer.Status, test_pred2$predicted)
knitr::kable(tab_test2)

	Churned	Joined	Stayed
Churned	361	41	142
Joined	37	83	7
Stayed	160	2	1279

round((sum(diag(tab_test2))/sum(tab_test2))*100,2)

## [1] 81.58

CustomerChurnPrediction

Ankur

17/08/2022

1. Objective

2. Overview

3. Data Description

3.1) Importing and merging the dataset

3.2) Inspecting the dataset

Summary

Strucure

Head

4. Map View

Map view of all observation

Count of users in each ZipCode

Percentage of User Churned in specific ZipCode

5. Data Cleaning and manipulation

Summary

Structure

Head

6. Data Split

Train Dataset

Test Dataset

7. Preliminary Data Analysis

Chi-square and Cramer’s V

Boxplot

MANOVA

Correlation

8. Multinomial Logistic Regression

Model Summary

Model Fit in Train Dataset & Accuracy

Model Fit in Test Dataset & Accuracy